@ebowwa/markdown-docs-scraper 1.1.0 → 1.2.0

package/src/index.ts

@@ -5,6 +5,7 @@
  * - Configurable llms.txt paths with fallbacks
  * - Custom URL patterns for different doc sites
  * - Works with any markdown documentation site
+ * - Uses full URLs from llms.txt directly
  */
 
 // ============================================================================
@@ -32,6 +33,8 @@ export interface ScraperOptions {
   tryDocsSubdomain?: boolean;
   /** Custom regex pattern to extract pages from llms.txt (must have 3 capture groups: title, fullUrl, path) */
   linkPattern?: RegExp;
+  /** Use full URLs from llms.txt directly (default: true for generic pattern) */
+  useDirectUrls?: boolean;
 }
 
 export interface ScraperResult {
@@ -40,11 +43,18 @@ export interface ScraperResult {
   duration: number;
 }
 
+/** Discovered page with full URL */
+interface DiscoveredPage {
+  category: string;
+  page: string;
+  fullUrl: string;  // The complete URL from llms.txt
+}
+
 /** Default pattern: matches /docs/en/ or /docs/ paths */
 const DEFAULT_LINK_PATTERN = /\[([^\]]+)\]\((https?:\/\/[^\s)]+\/docs\/(?:en\/)?([^)]+\.md))\)/g;
 
-/** Generic pattern: matches any .md links in llms.txt */
-const GENERIC_LINK_PATTERN = /\[([^\]]+)\]\((https?:\/\/[^\s)]+\/([^)]+\.md))\)/g;
+/** Generic pattern: matches any .md links - captures full path after domain */
+const GENERIC_LINK_PATTERN = /\[([^\]]+)\]\((https?:\/\/[^/]+\/([^\s)]+\.md))\)/g;
 
 // ============================================================================
 // UTILITY FUNCTIONS (Composable)
@@ -116,11 +126,12 @@ export class MarkdownDocsScraper {
       llmsPaths: options.llmsPaths || ["/llms.txt", "/docs/llms.txt"],
       tryDocsSubdomain: options.tryDocsSubdomain ?? true,
       linkPattern: options.linkPattern || GENERIC_LINK_PATTERN,
+      useDirectUrls: options.useDirectUrls ?? true,
     };
   }
 
   /**
-   * Build URL for a documentation page
+   * Build URL for a documentation page (fallback when no direct URL)
    */
   buildUrl(category: string, page: string): string {
     if (category) {
@@ -128,16 +139,19 @@ export class MarkdownDocsScraper {
     } else if (this.options.docsPath) {
       return `${this.options.baseUrl}${this.options.docsPath}/${page}.md`;
     } else {
-      // No docsPath (like Polymarket) - direct path
       return `${this.options.baseUrl}/${page}.md`;
     }
   }
 
   /**
-   * Download a single documentation page
+   * Download a page using either direct URL or built URL
    */
-  async downloadPage(category: string, page: string): Promise<DocPage | null> {
-    const url = this.buildUrl(category, page);
+  async downloadPage(pageInfo: DiscoveredPage): Promise<DocPage | null> {
+    // Use direct URL if available and useDirectUrls is enabled
+    const url = (this.options.useDirectUrls && pageInfo.fullUrl)
+      ? pageInfo.fullUrl
+      : this.buildUrl(pageInfo.category, pageInfo.page);
+
     const content = await fetchMarkdown(url);
 
     if (!content) {
@@ -148,8 +162,8 @@ export class MarkdownDocsScraper {
       url,
       title: extractTitle(content),
       content,
-      category,
-      pageName: page,
+      category: pageInfo.category,
+      pageName: pageInfo.page,
     };
   }
 
@@ -173,7 +187,6 @@ export class MarkdownDocsScraper {
 
         // Skip if already on docs/doc subdomain
         if (!hostname.startsWith("docs.") && !hostname.startsWith("doc.")) {
-          // Try docs.{domain}
           const docsDomain = hostname.replace(/^www\./, "");
           urls.push(`${url.protocol}//docs.${docsDomain}/llms.txt`);
           urls.push(`${url.protocol}//docs.${docsDomain}/docs/llms.txt`);
@@ -191,9 +204,11 @@ export class MarkdownDocsScraper {
    */
   private async fetchLlmsTxt(): Promise<{ content: string; url: string } | null> {
     const urls = this.getLlmsUrls();
+    console.log(`DEBUG: Trying URLs: ${urls.join(", ")}`);
 
     for (const llmsUrl of urls) {
       try {
+        console.log(`DEBUG: Fetching ${llmsUrl}...`);
         const response = await fetch(llmsUrl, {
           headers: {
             Accept: "text/plain",
@@ -201,13 +216,14 @@ export class MarkdownDocsScraper {
           },
         });
 
+        console.log(`DEBUG: Response status: ${response.status}`);
         if (response.ok) {
           const content = await response.text();
           console.log(`Found llms.txt at ${llmsUrl}`);
           return { content, url: llmsUrl };
         }
       } catch (error) {
-        // Try next URL
+        console.log(`DEBUG: Error: ${error}`);
         continue;
       }
     }
@@ -218,8 +234,8 @@ export class MarkdownDocsScraper {
   /**
    * Discover pages from llms.txt index
    */
-  async discoverPages(): Promise<Array<{ category: string; page: string }>> {
-    const pages: Array<{ category: string; page: string }> = [];
+  async discoverPages(): Promise<DiscoveredPage[]> {
+    const pages: DiscoveredPage[] = [];
 
     try {
       const llmsResult = await this.fetchLlmsTxt();
@@ -233,15 +249,20 @@ export class MarkdownDocsScraper {
       const { content } = llmsResult;
 
       // Use provided pattern or default
-      const regex = new RegExp(this.options.linkPattern.source, this.options.linkPattern.flags);
+      const pattern = this.options.linkPattern;
+      const regex = new RegExp(pattern.source, pattern.flags);
       let match;
 
+      // Debug: log pattern being used
+      console.log(`DEBUG: Using pattern: ${pattern.source}`);
+      console.log(`DEBUG: Content length: ${content.length}`);
+
       while ((match = regex.exec(content)) !== null) {
-        const url = match[2];
+        const fullUrl = match[2];  // The full URL from llms.txt
         const pagePath = match[3]; // The captured path group
 
         const { category, page } = parsePagePath(pagePath);
-        pages.push({ category, page });
+        pages.push({ category, page, fullUrl });
       }
 
       console.log(`Discovered ${pages.length} pages from llms.txt`);
@@ -273,7 +294,7 @@ export class MarkdownDocsScraper {
     for (let i = 0; i < pages.length; i += this.options.concurrency) {
       const batch = pages.slice(i, i + this.options.concurrency);
       const results = await Promise.allSettled(
-        batch.map((page) => this.downloadPage(page.category, page.page))
+        batch.map((page) => this.downloadPage(page))
       );
 
       results.forEach((result, index) => {
@@ -281,8 +302,11 @@ export class MarkdownDocsScraper {
         if (result.status === "fulfilled" && result.value) {
           downloaded.push(result.value);
         } else {
+          const url = (this.options.useDirectUrls && page.fullUrl)
+            ? page.fullUrl
+            : this.buildUrl(page.category, page.page);
           failed.push({
-            url: this.buildUrl(page.category, page.page),
+            url,
             error: result.status === "rejected" ? (result.reason as string) : "Not found",
           });
         }
@@ -316,7 +340,7 @@ export class MarkdownDocsScraper {
     for (let i = 0; i < pages.length; i += this.options.concurrency) {
       const batch = pages.slice(i, i + this.options.concurrency);
       const results = await Promise.allSettled(
-        batch.map((page) => this.downloadPage(page.category, page.page))
+        batch.map((page) => this.downloadPage({ ...page, fullUrl: "" }))
       );
 
       results.forEach((result, index) => {
@@ -350,7 +374,6 @@ export class MarkdownDocsScraper {
     const path = await import("path");
 
     for (const page of pages) {
-      // Use pageName if available, otherwise extract from URL
       const nameToUse = page.pageName || page.url.split("/").pop()?.replace(".md", "") || "untitled";
 
       const dir = page.category
@@ -369,12 +392,12 @@ export class MarkdownDocsScraper {
   /**
    * Get list of pages to scrape based on categories
    */
-  private getPagesToScrape(): Array<{ category: string; page: string }> {
-    const pages: Array<{ category: string; page: string }> = [];
+  private getPagesToScrape(): DiscoveredPage[] {
+    const pages: DiscoveredPage[] = [];
 
     for (const [category, pageList] of Object.entries(this.options.categories)) {
       for (const page of pageList) {
-        pages.push({ category, page });
+        pages.push({ category, page, fullUrl: "" });
       }
     }
 
@@ -424,6 +447,7 @@ export function claudeCodeOptions(outputDir: string): ScraperOptions {
     outputDir,
     concurrency: 10,
     tryDocsSubdomain: false,
+    useDirectUrls: false,  // Claude Code can use built URLs
   };
 }
 
@@ -437,9 +461,38 @@ export function polymarketOptions(outputDir: string): ScraperOptions {
     outputDir,
     concurrency: 10,
     tryDocsSubdomain: false,
+    useDirectUrls: true,  // Polymarket needs direct URLs
   };
 }
 
+// ============================================================================
+// SCRAPERS MODULE
+// ============================================================================
+
+/**
+ * Re-export scrapers module for composable scraper architecture.
+ * This provides a registry-based system for different scraper implementations.
+ */
+export {
+  // Types
+  type SourceType,
+  type SourceConfig,
+  type Scraper,
+  type ScrapeResult as ScraperModuleResult,
+  type DownloadResult,
+
+  // Scrapers
+  llmsTxtScraper,
+  githubRawScraper,
+  CLAUDE_CODE_PATTERN as SCRAPER_CLAUDE_CODE_PATTERN,
+  GENERIC_PATTERN as SCRAPER_GENERIC_PATTERN,
+
+  // Registry
+  registerScraper,
+  getScraper,
+  scrapeSource,
+} from "./scrapers/index";
+
 // ============================================================================
 // EXPORTS
 // ============================================================================

package/src/scrapers/github-raw.ts

@@ -0,0 +1,154 @@
+/**
+ * GitHub Raw Scraper
+ *
+ * Downloads markdown files directly from GitHub repositories via raw content URLs.
+ * Uses GitHub API to list files, then fetches each from raw.githubusercontent.com
+ */
+
+import type { Scraper, SourceConfig, ScrapeResult, DownloadResult } from "./types";
+
+// ============================================================================
+// GITHUB API TYPES
+// ============================================================================
+
+interface GitHubContent {
+  name: string;
+  path: string;
+  download_url: string;
+  type: string;
+}
+
+// ============================================================================
+// GITHUB RAW SCRAPER
+// ============================================================================
+
+export const githubRawScraper: Scraper = {
+  type: "github-raw",
+
+  async scrape(config: SourceConfig): Promise<ScrapeResult> {
+    const startTime = Date.now();
+    const downloaded: DownloadResult[] = [];
+    const failed: Array<{ url: string; error: string }> = [];
+
+    if (!config.github?.repo) {
+      throw new Error(`GitHub source "${config.name}" missing github.repo config`);
+    }
+
+    // Get list of markdown files from GitHub API
+    const files = await fetchGitHubMarkdownFiles(
+      config.github.repo,
+      config.docsPath.replace(/^\//, "")
+    );
+
+    // Download each file
+    for (const file of files) {
+      const content = await fetchGitHubRawContent(config.github.repo, file.path);
+
+      if (content) {
+        downloaded.push({
+          success: true,
+          path: file.name,
+          title: extractTitle(content) || file.name.replace(".md", ""),
+        });
+
+        // Save the file
+        await saveFile(config.outputDir, file.name, content);
+      } else {
+        failed.push({
+          url: `https://raw.githubusercontent.com/${config.github.repo}/main/${file.path}`,
+          error: "Failed to fetch content",
+        });
+      }
+    }
+
+    return {
+      downloaded,
+      failed,
+      duration: Date.now() - startTime,
+    };
+  },
+};
+
+// ============================================================================
+// GITHUB API FUNCTIONS
+// ============================================================================
+
+/**
+ * Fetch list of markdown files from GitHub repo directory
+ */
+async function fetchGitHubMarkdownFiles(
+  repo: string,
+  path: string
+): Promise<GitHubContent[]> {
+  const url = `https://api.github.com/repos/${repo}/contents/${path}`;
+
+  const response = await fetch(url, {
+    headers: {
+      Accept: "application/vnd.github.v3+json",
+      "User-Agent": "@ebowwa/markdown-docs-scraper",
+    },
+  });
+
+  if (!response.ok) {
+    throw new Error(`GitHub API error: ${response.status} ${response.statusText}`);
+  }
+
+  const contents: GitHubContent[] = await response.json();
+
+  // Filter for markdown files only
+  return contents.filter(
+    (item) => item.type === "file" && item.name.endsWith(".md")
+  );
+}
+
+/**
+ * Download markdown content from GitHub raw URL
+ */
+async function fetchGitHubRawContent(
+  repo: string,
+  path: string
+): Promise<string | null> {
+  const url = `https://raw.githubusercontent.com/${repo}/main/${path}`;
+
+  try {
+    const response = await fetch(url, {
+      headers: {
+        Accept: "text/plain",
+        "User-Agent": "@ebowwa/markdown-docs-scraper",
+      },
+    });
+
+    if (!response.ok) {
+      return null;
+    }
+
+    return await response.text();
+  } catch (error) {
+    console.error(`Error fetching ${url}:`, error);
+    return null;
+  }
+}
+
+/**
+ * Extract title from markdown content
+ */
+function extractTitle(markdown: string): string | null {
+  const titleMatch = markdown.match(/^#\s+(.+)$/m);
+  return titleMatch ? titleMatch[1].trim() : null;
+}
+
+/**
+ * Save file to disk
+ */
+async function saveFile(
+  outputDir: string,
+  filename: string,
+  content: string
+): Promise<void> {
+  const fs = await import("fs/promises");
+  const path = await import("path");
+
+  const outputPath = path.join(outputDir, filename);
+  await fs.mkdir(path.dirname(outputPath), { recursive: true });
+  await fs.writeFile(outputPath, content, "utf-8");
+}

package/src/scrapers/index.ts

@@ -0,0 +1,16 @@
+/**
+ * Scrapers Module
+ *
+ * Composable scraper architecture for multiple documentation source types.
+ * This module provides a registry-based system for different scraper implementations.
+ */
+
+// Types
+export type { SourceType, SourceConfig, Scraper, ScrapeResult, DownloadResult } from "./types";
+
+// Scrapers
+export { llmsTxtScraper, CLAUDE_CODE_PATTERN, GENERIC_PATTERN } from "./llms-txt";
+export { githubRawScraper } from "./github-raw";
+
+// Registry
+export { registerScraper, getScraper, scrapeSource } from "./registry";

package/src/scrapers/llms-txt.ts

@@ -0,0 +1,101 @@
+/**
+ * LLMS-TXT Scraper
+ *
+ * Scrapes documentation sites that provide llms.txt index files.
+ * Uses the core MarkdownDocsScraper under the hood.
+ */
+
+import { scrapeMarkdownDocs, type DocPage } from "../index";
+import type { Scraper, SourceConfig, ScrapeResult, DownloadResult } from "./types";
+
+// ============================================================================
+// URL PATTERNS
+// ============================================================================
+
+/** Pattern for Claude Code docs: /docs/en/page.md */
+export const CLAUDE_CODE_PATTERN = /\[([^\]]+)\]\((https?:\/\/[^\s)]+\/docs\/en\/([^)]+\.md))\)/g;
+
+/** Pattern for generic docs: any domain/path.md */
+export const GENERIC_PATTERN = /\[([^\]]+)\]\((https?:\/\/[^/]+\/([^\s)]+\.md))\)/g;
+
+// ============================================================================
+// LLMS-TXT SCRAPER
+// ============================================================================
+
+export const llmsTxtScraper: Scraper = {
+  type: "llms-txt",
+
+  async scrape(config: SourceConfig): Promise<ScrapeResult> {
+    const options = getScraperOptions(config);
+    const result = await scrapeMarkdownDocs(options);
+
+    // Convert DocPage[] to DownloadResult[]
+    const downloaded: DownloadResult[] = result.downloaded.map((page: DocPage) => {
+      const category = page.category || "";
+      const filename = `${page.pageName || "untitled"}.md`;
+      const path = category ? `${category}/${filename}` : filename;
+
+      return {
+        success: true,
+        path,
+        title: page.title,
+      };
+    });
+
+    return {
+      downloaded,
+      failed: result.failed,
+      duration: result.duration,
+    };
+  },
+};
+
+// ============================================================================
+// OPTIONS BUILDER
+// ============================================================================
+
+/**
+ * Get scraper options based on source configuration
+ */
+function getScraperOptions(config: SourceConfig) {
+  const baseOptions = {
+    baseUrl: config.baseUrl,
+    docsPath: config.docsPath,
+    outputDir: config.outputDir,
+    concurrency: 10,
+    useLlms: true,
+    tryDocsSubdomain: false,
+  };
+
+  // Source-specific options
+  if (config.name === "Claude Code") {
+    return {
+      ...baseOptions,
+      llmsPaths: ["/docs/llms.txt"],
+      linkPattern: CLAUDE_CODE_PATTERN,
+    };
+  }
+
+  if (config.name === "Polymarket") {
+    return {
+      ...baseOptions,
+      llmsPaths: ["/llms.txt"],
+      linkPattern: GENERIC_PATTERN,
+    };
+  }
+
+  if (config.name === "Bun") {
+    return {
+      ...baseOptions,
+      llmsPaths: ["/docs/llms.txt", "/llms.txt"],
+      linkPattern: GENERIC_PATTERN,
+    };
+  }
+
+  // Default: use provided llmsTxtPath or try common paths
+  return {
+    ...baseOptions,
+    llmsPaths: config.llmsTxtPath ? [config.llmsTxtPath] : ["/llms.txt", "/docs/llms.txt"],
+    linkPattern: config.linkPattern || GENERIC_PATTERN,
+  };
+}

package/src/scrapers/registry.ts

@@ -0,0 +1,55 @@
+/**
+ * Scraper Registry
+ *
+ * Maps source types to scraper implementations.
+ * Allows registering new scrapers and looking them up by type.
+ */
+
+import type { Scraper, SourceType, SourceConfig, ScrapeResult } from "./types";
+import { llmsTxtScraper } from "./llms-txt";
+import { githubRawScraper } from "./github-raw";
+
+// ============================================================================
+// SCRAPER REGISTRY
+// ============================================================================
+
+/** Registry of all available scrapers keyed by type */
+const scrapers: Map<SourceType, Scraper> = new Map();
+
+/**
+ * Register a scraper implementation
+ */
+export function registerScraper(scraper: Scraper): void {
+  scrapers.set(scraper.type, scraper);
+}
+
+/**
+ * Get a scraper by type
+ */
+export function getScraper(type: SourceType): Scraper | undefined {
+  return scrapers.get(type);
+}
+
+/**
+ * Scrape a source using the appropriate scraper
+ */
+export async function scrapeSource(config: SourceConfig): Promise<ScrapeResult> {
+  const scraper = scrapers.get(config.sourceType);
+
+  if (!scraper) {
+    throw new Error(`No scraper registered for type: ${config.sourceType}`);
+  }
+
+  return scraper.scrape(config);
+}
+
+// ============================================================================
+// DEFAULT REGISTRATIONS
+// ============================================================================
+
+// Register built-in scrapers
+registerScraper(llmsTxtScraper);
+registerScraper(githubRawScraper);
+
+// Export scrapers for direct access if needed
+export { llmsTxtScraper, githubRawScraper };

package/src/scrapers/types.ts

@@ -0,0 +1,79 @@
+/**
+ * Scraper Types
+ *
+ * Core types for the composable scraper architecture.
+ * These types define the interface that all scrapers must implement.
+ */
+
+// ============================================================================
+// SOURCE TYPES
+// ============================================================================
+
+/** Supported documentation source types */
+export type SourceType = "llms-txt" | "github-raw";
+
+// ============================================================================
+// SCRAPER INTERFACE
+// ============================================================================
+
+/** Result from scraping a source */
+export interface ScrapeResult {
+  downloaded: DownloadResult[];
+  failed: Array<{ url: string; error: string }>;
+  duration?: number;
+}
+
+/** Individual download result */
+export interface DownloadResult {
+  success: boolean;
+  path: string;
+  title?: string;
+}
+
+/** Scraper interface - all scrapers must implement this */
+export interface Scraper {
+  /** Source type identifier */
+  type: SourceType;
+
+  /** Scrape documentation from a source */
+  scrape(config: SourceConfig): Promise<ScrapeResult>;
+}
+
+// ============================================================================
+// SOURCE CONFIG
+// ============================================================================
+
+/** Source configuration */
+export interface SourceConfig {
+  /** Display name */
+  name: string;
+
+  /** Source type - determines which scraper to use */
+  sourceType: SourceType;
+
+  /** Base URL for the documentation */
+  baseUrl: string;
+
+  /** Path to docs (e.g., /docs, /docs/en) */
+  docsPath: string;
+
+  /** Output directory for downloaded docs */
+  outputDir: string;
+
+  /** Output directory for daily reports */
+  reportDir: string;
+
+  /** llms.txt path (for llms-txt sources) */
+  llmsTxtPath?: string;
+
+  /** Custom link pattern for llms.txt parsing */
+  linkPattern?: RegExp;
+
+  /** GitHub config (for github-raw sources or GitHub API data) */
+  github?: {
+    repo: string;
+    includeCommits: boolean;
+    includeReleases: boolean;
+    includePRs: boolean;
+  };
+}